Genotyping of Bartonella Bacteria and Their Animal Hosts: Current Status and Perspectives

543

SPECIAL ISSUE REVIEW Genotyping of Bartonella bacteria and their animal hosts: current status and perspectives

M. KOSOY1*†,C.MCKEE1,2†, L. ALBAYRAK3 and Y. FOFANOV3 1 Division of Vector-Borne Diseases, Centers for Disease Control and Prevention, Fort Collins, CO 80521, USA 2 Department of Biology, Colorado State University, Fort Collins, CO 80523, USA 3 Department of Pharmacology, University of Texas Medical Branch, Galveston, TX 77555, USA

(Received 8 May 2017; revised 7 June 2017; accepted 15 June 2017; ﬁrst published online 2 August 2017)

SUMMARY

Growing evidence demonstrates that bacterial species diversity is substantial, and many of these species are pathogenic in some contexts or hosts. At the same time, laboratories and museums have collected valuable animal tissue and ectoparasite samples that may contain substantial novel information on bacterial prevalence and diversity. However, the identification of bacterial species is challenging, partly due to the difficulty in culturing many microbes and the reliance on molecular data. Although the genomics revolution will surely add to our knowledge of bacterial systematics, these approaches are not accessible to all researchers and rely predominantly on cultured isolates. Thus, there is a need for comprehensive molecular analyses capable of accurately genotyping bacteria from animal tissues or ectoparasites using common methods that will facilitate large-scale comparisons of species diversity and prevalence. To illustrate the challenges of genotyping bacteria, we focus on the genus Bartonella, vector-borne bacteria common in mammals. We highlight the value and limitations of commonly used techniques for genotyping bartonellae and make recommendations for researchers interested in studying the diversity of these bacteria in various samples. Our recommendations could be applicable to many bacterial taxa (with some modifications) and could lead to a more complete understanding of bacterial species diversity.

Key words: barcoding, bacterial species, Bartonella, genomics, genotyping, parasite ecology, parasite–host association.

INTRODUCTION Although bacterial DNA analysis has been readily accepted for measuring the assembly, diversity and Since the proposal of DNA barcoding by Hebert distribution of entire microbial communities in et al. (2003) as a new methodology for identification different environments (DeLong and Pace, 2001), of biological species, it has been utilized on a wide application of universal barcoding approaches to variety of taxa for the purposes of identifying museum specimens, evaluation of population and bacteria may face challenges unique to prokaryotes. community diversity, discovery of cryptic species, First, many microbial species are challenging to and other forensic applications. Barcoding was culture by traditional methods (Schloss and adapted for species-level identification by recovery Handelsman, 2004), which will curtail a microbiolo- ’ of short DNA sequences from a specific genome gist s ability to characterize these species beyond fragment and has been applied widely in processing molecular approaches. Second, there are still discus- and identifying animal and plant tissues. Attempts sions about the precise definition of bacterial species have also been made to apply this barcoding para- and the numerous, competing methods used to char- digm to other eukaryote taxa and some microscopic acterize them (Konstantinidis et al. 2006; Fraser organisms (Blaxter, 2016). For example, the et al. 2007). Finally, the phylogenetic diversity nuclear ribosomal internal transcribed spacer (ITS) observed in microbes is far more complex than in region has been recognized as a potentially universal eukaryotes (Hug et al. 2016), with predictions of DNA barcode marker for fungi (Bellemain et al. global bacterial diversity ranging from 107 to 1012 2010; Schoch et al. 2012). Nevertheless, acceptance species (Curtis et al. 2002; Schloss and of DNA barcoding for the identification of organ- Handelsman, 2004; Dykhuizen, 2005; Locey and isms is controversial among some taxonomists Lennon, 2016). We argue that although our ability because of fears that a universal, DNA-based to measure bacterial diversity at a massive scale has approach in identification of species will replace increased in recent decades, the identification of bac- traditional methods (Lebonah et al. 2014). terial species is complex and may not be amenable to one universal approach. * Corresponding author: Division of Vector-Borne Diseases, Centers for Disease Control and Prevention, Meanwhile, microbiologists are faced with the Fort Collins, CO 80521, USA. E-mail: [email protected] dilemma of processing the large bank of specimens † These authors contributed equally to this work. sitting in laboratory freezers and museum collections

Parasitology (2018), 145, 543–562. © Cambridge University Press 2017. This is an Open Access article, distributed under the terms of the Creative Commons Attribution licence (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted re-use, distribution, and reproduction in any medium, provided the original work is properly cited. Downloaded fromdoi:10.1017/S0031182017001263 https://www.cambridge.org/core. IP address: 170.106.35.234, on 27 Sep 2021 at 11:08:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0031182017001263 M. Kosoy and others 544

across the globe, containing uncharacterized bacterial investigations in an effort to define the source of species and potential pathogens. These specimens human and animal diseases. The importance of may include irreplaceable tissue samples from rare tracing pathogenic bacteria in environment was or hard to reach animal species, samples from highlighted by threats presented by select biological human subjects or enormous collections of arthropod agents, particularly anthrax (Keim et al. 2008). In a ectoparasites. How should microbiologists approach more common situation, extensive databases con- assessing the diversity of bacteria parasitizing these taining sequences of multiple bacterial strains can diverse animals in a consistent manner that will facili- provide irreplaceable information for a comparison tate comparative epidemiological, evolutionary and of gene sequences between a presumptive human ecological analyses? For the remainder of this pathogen and potential zoonotic sources. As an review, we will focus on the diverse genus Bartonella example related to the study of bartonellae, a (Alphaproteobacteria: Rhizobiales) to make specific recent case of lymphadenopathy in Tbilisi, Georgia recommendations for the genotyping of these bacteria was linked to infected rats (Kandelaki et al. 2016). that could be applicable to a wider array of bacterial This was only possible because of the accurate taxa. Bartonellae are facultative, fastidious, intracellu- genetic characterization of related Bartonella lar bacteria commonly found in many taxonomically strains from commensal rats in Israel by Harrus diverse mammalian species globally (Kosoy et al. et al. (2009). Similar connections have been made 2012). Bartonellae are hypothesized to be transmitted between human cases of myocarditis and meningitis (some possibly harboured) by a variety of arthropod in the USA and Bartonella genotypes found in vectors, including ticks, mites, lice, fleas, flies and ground squirrels (Kosoy et al. 2003; Osikowicz other insects (Billeter et al. 2008;Tsaiet al. 2011). et al. 2016). Genotyping bartonellae and other infec- Given the high prevalence and broad host range of tious bacteria using consistent molecular approaches Bartonella species, we expect a wide variety of that generate a common repertoire of gene sequences Bartonella species and genotypes to be present in will surely increase the feasibility and frequency of animal tissues collected during field investigations or these comparisons. archived in museum and laboratory collections. Furthermore, this genus exemplifies many of the chal- Bartonellae as a popular tool for ecological studies lenges to characterizing bacterial species, including the limitations of 16S ribosomal RNA (rRNA) Bartonella species span the symbiont–pathogen con- sequencing (Kosoy et al. 2012), challenges with cul- tinuum (Segers et al. 2017) and are an extremely turing and frequent homologous recombination diverse group of bacteria, especially in rodents and among genomic loci (Berglund et al. 2010;Chaloner bats (Lei and Olival, 2014). Moreover, these verte- et al. 2011; Paziewska et al. 2011, 2012;Guyet al. brate host–arthropod vector–Bartonella systems 2012;Buffet et al. 2013a;Baiet al. 2015a). Our ana- appear to be globally distributed and phylogenetic- lyses in this review will be structured as follows: (1) ally complex. Such features make these tripartite highlight the value of accurate bacterial genotyping systems a popular tool for ecological comparative for epidemiological and ecological research, (2) analyses (Buffet et al. 2013b; Klangthong et al. address the challenge of identifying a bacterial 2015; Brook et al. 2017). There are some ecological species in animal tissues using a short sequence of projects where identification of Bartonella is not one or a few selected genomic fragments, (3) analyse essential and where a priority is given to estimation the literature on molecular identification of bartonel- of Bartonella prevalence in animal populations lae in different animal tissues and from diverse without identification of the species (Bai et al. animal taxa, (4) compare genetic markers used for 2009; Young et al. 2014). However, most ecological genotyping bartonellae with and without culturing and epidemiological studies require accurate iden- and (5) evaluate the phylogenetic resolution of candi- tification of specific bacterial species and/or geno- date loci based on analysis of genomes available for types. The level of discrimination between multiple Bartonella species. In the Discussion, we obtained strains or genotypes depends on the objec- will make specific recommendations for consistent tives of the studies. In most situations, the investiga- methods for genotyping bartonellae that will facilitate tors prefer to report bacteria at the species level or comparative studies. compare sequence identity with a specific Bartonella type strain. There is however a potential pitfall in reporting PCR-positive samples without NEED FOR ACCURATE GENOTYPING OF sequencing of positive products. Some ecologists BARTONELLA SPECIES IN ANIMAL TISSUES interested in using simple techniques for estimating prevalence of common animal infection may not be Importance of genotyping bartonellae in epidemiology aware that the primers and real-time PCR probes and for biological threat preparation selected for molecular detection of Bartonella Genotyping of pathogenic zoonotic bacteria can DNA may not always be specific for the genus be a very important part of epidemiological Bartonella (Maggi and Breitschwerdt, 2005). In the

Downloaded from https://www.cambridge.org/core. IP address: 170.106.35.234, on 27 Sep 2021 at 11:08:10, subject to the Cambridge Core terms of use, available at https://www.cambridge.org/core/terms. https://doi.org/10.1017/S0031182017001263 Genotyping of Bartonella bacteria and their animal hosts: current status and perspectives 545

absence of sequence data, reporting PCR-positive agar. Thus, a strictly culture-based assessment may samples alone may overestimate Bartonella preva- be severely biased towards cultivable strains. lence in such ecological studies. Therefore, we advocate that studies of Bartonella prevalence, and ideally The 16S ribosomal RNA sequencing and the problem of all surveys of infectious bacteria, should adhere to identification of Bartonella species the standard of reporting only sequence-positive samples. Besides the clarity of the results, assessing Assessing bacterial diversity using 16S rRNA the diversity and prevalence of Bartonella species sequences has become a very popular technique, in ecological studies with sequence-based especially with the advent of high-throughput approaches will facilitate comparisons of bacterial sequencing instruments (e.g. Roche 454, Illumina prevalence and diversity across various spatial and MiSeq/HiSeq and Ion Torrent). This approach temporal scales and among host species and commu- has been able to uncover an enormous diversity of nities. These analyses are necessary for a deeper bacterial and archaeal taxa, some of it consisting of understanding of the ecology and evolution of heretofore uncultured microbial ‘dark matter’ Bartonella species and other infectious bacteria. (Rinke et al. 2013; Saw et al. 2015), in environments ranging from mammalian guts and feces (Manichanh et al. 2006; Bittar et al. 2014) and parasitic arthro- CURRENT APPROACHES TO GENOTYPING pods (Qiu et al. 2014; Razzauti et al. 2015)to BARTONELLAE marine habitats (Logares et al. 2014). Modern microbiologists rely heavily on molecular Nevertheless, there is substantial evidence that techniques and associated thresholds to assess the 16S rRNA sequencing (and its high-throughput diversity of bacteria in various environments and applications) may be inadequate for accurately iden- classify new species. Sequencing of the 16S rRNA tifying bacterial species and may not be sensitive gene (Woese and Fox, 1977) and DNA–DNA enough to recover the complex evolutionary histor- hybridization experiments (Wayne et al. 1987) are ies of many microbial species. Assessment of micro- two methods capable of delineating bacterial bial taxonomic diversity using 16S rRNA sequences species that were developed fairly early, yet in the commonly follows a threshold of 97% sequence iden- intervening years, the limitations of these approach tity to delineate operational taxonomic units, which have become more clear (Konstantinidis et al. can obscure the distinction between closely related 2006). We will highlight some of these limitations species and even genera (e.g. Escherichia and below, and where applicable, make specific connec- Shigella). Many authors have determined that tions to the study of bartonellae. Bartonella species exhibit very high levels of 16S rRNA gene sequence similarity (Birtles and Raoult, 1996). Comparing sequences of 17 Is bacterial isolation a necessary step for genotyping of Bartonella species and subspecies, La Scola et al. bartonellae? (2003) reported the lowest discriminatory power The isolation of Bartonella bacteria from infected (99·7%) and highest interspecies similarity (99·8%) animals is the preferred method for the diagnosis for 16S rRNA, making this genetic locus an ineffec- and characterization of species (Gutiérrez et al. tive tool for the systematic classification of related 2017). Gutiérrez et al. (2017) discuss various bacterial species. methods that are successful in culturing bartonellae and make recommendations for particular sample Metagenomics of microbial communities and needs for types (tissue vs ectoparasite). However, one persist- Bartonella genotyping ent challenge is that many microbes will be challenging to isolate with known culturing techniques. We While reporting the low discriminatory power of do acknowledge that cultured isolates are crucially 16S rRNA for identification of Bartonella species, important to the identification of bacterial species La Scola et al. (2003) acknowledged that this locus and obtaining a culture will facilitate all varieties of is still reliable for differentiation of all Bartonella morphological, biochemical and genetic analyses. species from Brucella species (94% similarity), the There are projects already underway to sequence genus taxonomically closest to Bartonella. This fact the whole genome of all known bacterial-type can justify the application of ribosomal primers for strains (Kyrpides et al. 2014), and these data will identification of bartonellae as components of micro- no doubt expand our knowledge of bacterial diver- bial communities using 16S rRNA amplicon sity and genomic architecture. For bartonellae spe- sequencing. Few surveys based on metagenomic cifically, culturing can be very time-consuming due evaluation of rodent-associated bacteria, including to the slow growth of the bacteria (which can be bartonellae, were conducted recently (Razzauti complicated by overgrowth of other contaminating et al. 2015; Galan et al. 2016; Koskela et al. 2017). bacteria) and may not be able to detect some These studies highlight the utility of metagenomic species that do not grow quickly on standard blood techniques; however, they are not without their

own limitations regarding the distinction among known zoonotic agent in humans (Welch et al. 1999; related bacterial species and potential sequence Bai et al. 2012). amplification biases. Banskar et al. (2016) used 16S metagenomic Razzauti et al. (2015) compared two next-gener- sequencing (Ion Torrent) to investigate the fecal ation sequencing approaches (transcriptome RNA microbiome of Rousettus leschenaultii bats in India. sequencing and 16S metagenomics) according to They found a high abundance of Proteobacteria in their ability to survey multiple bacteria in rodent some of the samples, which contains a large populations in the French Ardennes region. number of pathogenic genera, including Among vector-borne bacteria, Bartonella was the Bartonella. However, the authors claim to have most prevalent (>5 reads in 89% of the rodents by detected Bartonella henselae in two of the bat 16S sequencing). The authors acknowledged that samples, which is highly unlikely given the strong an important limitation of these approaches is the association of B. henselae with cats. This misidentifi- accuracy of the taxonomic assignation. RNA cation is most likely due to the authors’ use of a 97% sequencing allowed taxonomic classification at the sequence identity threshold, which is insufficient to species level, while 16S metagenomics classification distinguish among Bartonella species. Dietrich was generally restricted to the genus level. et al. (2017) also applied 16S metagenomic sequen- Analysing this problem, Razzauti et al. (2015) cing (Illumina MiSeq) to characterize the micro- stressed the point that the 16S rRNA gene is biome in saliva, urine and feces from four species difficult to sequence in its totality because of the of insectivorous bats from South Africa. Similarly size (∼1550 bp) using current high-throughput to Banskar et al. the authors found a high abundance sequencing methods. The method proposed by of Proteobacteria in bat feces, but also in saliva and Miller et al. (2011) allows to assembly steps, but is urine. Sequences mapping to Bartonella were not frequently used because of the increased experi- found predominantly in feces, but also to some mental complexity and cost. Instead, a portion of the extant in saliva and urine. No attempts were made 16S rRNA gene is usually amplified using specific to identify the specific Bartonella species found in sets of universal primers. The nine hypervariable these samples; however, as we have discussed (V) regions of the 16S rRNA gene differ between above, this would likely not be possible using only species, and depending on the V region chosen, 16S sequences. one can discriminate some species but not others. In review, the utility of 16S sequencing will In their paper, Razzauti et al. (2015) have also largely depend on the questions investigators wish reported an interesting observation about a large to pursue, and the scale of phylogenetic resolution difference in the relative abundance of Bartonella needed to answer such questions. If investigators reads detected by the 16S MiSeq (95%) vs RNA wish to assess bacterial diversity in specimens at sequencing (<1%). the genus level or higher, 16S metagenomics would Galan et al. (2016) investigated the potential for be an excellent approach. As we have reviewed recent developments in 16S rRNA-based high- above, below the genus level however, this gene throughput sequencing (Illumina MiSeq) to facili- will not be sufficient to accurately distinguish tate the multiplexing of urban rodents in West among related species. Depending on the focus of Africa. This study reported significant difference in the study, investigators could then target a few Bartonella prevalence between rodent species genera of interest for characterization with more dis- varying from 0·5% in Mus musculus to 79% in criminating genetic loci (André et al. 2017). For Mastomys natalensis. Praising advances in this example, investigators may target genera with high screening strategy, the authors admit that 16S abundance in the 16S dataset that may contain rRNA amplicon sequencing based on a short pathogenic species. This approach has been used sequence did not yield results sufficiently high in with success recently to describe Bartonella species resolution to distinguish between Bartonella in bats (Veikkolainen et al. 2014; Wilkinson et al. species (Galan et al. 2016). Another metagenomic 2016). evaluation of bacteria in voles from Finland (Koskela et al. 2017) reported commonality of DNA–DNA hybridization and its limitations for Bartonella species in the voles, although identifica- identification of bacteria tion of the species was not clear. André et al. (2017) used 16S metagenomics to investigate the The DNA–DNA hybridization experiments used by liver microbiome of Peromyscus leucopus mice in Wayne et al. (1987) represented a potentially more Canada, finding no difference between the micro- robust approach for delineating bacterial species biome assemblages of mouse genotypes separated using the whole genome. A threshold of 70% hybrid- by the Saint Lawrence River. In contrast to the ization has been used as the ‘gold standard’ criterion other studies above, the authors used an additional for distinguishing new bacterial species (Tindall marker (16S–23S intergenic spacer, ITS) to identify et al. 2010); however, this technique requires the all of the Bartonella species as B. vinsonii arupensis,a use of cultured isolates, specialized equipment and

multiple confirmatory tests due to variation across Gene-sequence-based paradigm for identification of experimental runs. Alternative genome-wide dis- Bartonella isolates tance measures included average nucleotide identity There are existing methods, particularly multi-locus (Konstantinidis and Tiedje, 2005; Konstantinidis sequence typing (MLST; Stackebrandt et al. 2002), et al. 2006; Goris et al. 2007; Richter and Rosselló- which can balance the tradeoffs of culturing bias, Móra, 2009) and digital DNA–DNA hybridization phylogenetic resolution, homologous recombination (Auch et al. 2010; Meier-Kolthoff et al. 2014a, b). and gene conservation across species. MLST of One considerable advantage of these techniques is house-keeping genes (i.e. genes under stabilizing that they do not necessarily require cultured isolates selection encoding metabolic functions) remains a and can be calculated from draft genomes assembled powerful technique that can be used on uncultured from metagenome and transcriptome sequencing. bacteria to detect evidence of mixed infections and/ However, as noted above, these high-throughput or homologous recombination, provide sufficient techniques are currently not accessible to very phylogenetic resolution for the delineation of bacter- many research groups and will not have much ial species, and will provide consistency in the usage utility unless investigators have a bacterial isolate of genetic loci that can facilitate global assessments or a draft genome. of parasitic bacterial diversity. Using such an approach, La Scola et al. (2003) Recombination as an important complication for compared the similarities of seven genetic loci genotyping among the 17 species and subspecies of genus Bartonella. This comparison led to both the defini- Another important issue that can arise when tion of similarity values that discriminated attempting to genotype a bacterial strain is that sep- Bartonella at the species level and assessment of the arate genes may indicate the presence of different relative discriminatory power of each gene examined. species. These conflicts arise due to lateral gene The gltA, groEL, rpoB and ftsZ genes, and ITS all transfer (LGT) among bacteria, either directly have good discriminating power ranging from 92·6 through conjugation or indirectly via phage- to 94·4%. Overall, two genes (rpoB and gltA) were mediated transduction or transformation by uptake found to be the most potent markers for demarcation of free DNA in the environment. LGT is the pre- of Bartonella species (La Scola et al. 2003). This dominant mechanism by which bacteria acquire paper was very influential for characterization of antibiotic resistance genes and can be an important Bartonella cultures and defining their status as a part of bacterial evolution (Vos, 2009). species. Many studies have now used MLST Homologous recombination is a specific form of approaches to characterize Bartonella genotypes and LGT whereby homologous genes of a donor species, and based on the available methods, some genome replace the gene variant in the recipient form of multi-locus sequencing appears to be the genome. Homologous recombination is a common most viable method available to most researchers. feature among some bacterial species (Vos and Didelot, 2008) and even among distantly related bacteria (Hanage et al. 2006), thus severely compli- Confirmation of species status based on comparison of cating phylogenetic inference. This problem has gene profiles among related species been documented in several studies of Bartonella strains from cats, rodents and bats based on Characterizing the novel Bartonella species (B. melo- sequencing multiple protein-coding loci (Berglund phagi) isolated from sheep blood and sheep keds, et al. 2010; Chaloner et al. 2011; Paziewska et al. Kosoy et al. (2016) reported presence of 183 genes 2011, 2012;Guyet al. 2012;Buffet et al. 2013a; specific for this species, being absent in genomes of Bai et al. 2015a) and have provided valuable infor- other Bartonella species associated with ruminants mation about the mechanisms that generate to support their argument for the separation of this Bartonella diversity and the gene flow among co- bacterial species from species of other ruminant- occurring species. We note here that these studies associated Bartonella species. The authors identified have been limited to cultured strains. As we will that out of the 1338 genes, the number of homolo- discuss later in the paper, attempts to genotype gous but unique genes was estimated to be 1274, bartonellae from genomic DNA extracted from out of which 156 genes appeared to be specificto whole blood, tissue or from ectoparasites may be B. melophagi and absent in any of the 21 reference further complicated by the presence of multiple genomes Bartonella species. Comparison of the Bartonella species in the sample. However, sequen- gene profile of B. melophagi with related Bartonella cing multiple loci will clarify if recombination or species associated with ruminants (B. bovis and B. multiple species are present and phylogenetic con- schoenbuchensis) demonstrated that 183 genes were cordance among sequenced loci can be sufficient to present only in the genome of B. melophagi, while describe a potentially novel Bartonella species or 1027 genes present in one or more copies in each subspecies. genome were conserved between these three

bacterial strains. The remaining 27 genes were culturing, cloning sequences into vectors before present in B. melophagi and absent in related sequencing, or deep sequencing approaches may species (B. bovis and B. schoenbuchensis), but found differentiate these possible scenarios. In some in at least one of the other Bartonella species. This cases, multiple peaks may be visible in chromato- analysis indicates that even among related bacterial grams of sequences, so cloning would be useful in species, genomes can be very flexible in gene these cases, but not all cases of multiple infections content, and can be useful criterion for describing show this pattern, probably due to varying abun- novel species (Konstantinidis and Tiedje, 2005; dances of DNA that are not detected in the consen- Konstantinidis et al. 2006). sus sequence reads. Even with these known limitations, this multi- locus sequencing approach has been used recently Applying multi-locus approaches to detection and to detect and characterize Bartonella genotypes sequencing without cultured isolates from bats, rodents and carnivores. Lilley et al. Due to the challenges of culturing bartonellae, inves- (2015) used a combination of rpoB and gltA tigators may choose to characterize Bartonella infec- sequences to characterize a novel Bartonella species tions in animal samples directly from extracted (Candidatus B. hemsundetiensis)inMyotis daubento- DNA. Despite the convenience of such an approach, nii bats from Finland. Similar Bartonella species it has its own challenges. For instance, some primers have subsequently been cultured and characterized used to amplify genetic loci may be insufficiently from related insectivorous bats in the Republic of sensitive to amplify Bartonella DNA from some Georgia (Urushadze et al. 2017). Martin-Alonso animal tissues. The presence of PCR inhibitors et al. (2016) used multiple loci (ITS, gltA and that carry through from blood, tissue or the extrac- rpoB) to detect Bartonella infections in rodent tion process may also interfere with detection. The species from Benin. Based on these markers, the use of nested PCR reactions may be able to overcome authors describe a distinct Bartonella species some of these deficiencies among sensitivity among (Candidatus B. mastomydis) from M. natalensis. loci for detection and may not require much add- Sequences very similar to this candidate species itional primer design. For instance, protocols exist had previously been acquired from related rodent for amplifying gltA and ftsZ sequences using species in Ethiopia. The authors also reported the nested reactions with known primer sets (Norman presence of multiple peaks in their sequencing et al. 1995; Birtles and Raoult, 1996; Zeaiter et al. results, so they used cloning to distinguish the coin- 2002; Colborn et al. 2010; Gundi et al. 2012a). fections. However, there were additional conflicts Another challenge with amplification directly between gltA and rpoB sequences for some samples from extracted DNA is the potential presence of that did not show multiple sequence peaks, with multiple bacterial species in the sample, and coinfec- one locus indicating the presence of Bartonella eliza- tions of multiple Bartonella genotypes in one animal bethae and the other indicating Bartonella tribocorum are not uncommon based on culturing. When mul- (Martin-Alonso et al. 2016). The authors hypothe- tiple species are present in a sample, the abundance size that these conflicts may have arisen by recom- of their DNA may vary in the sample and even bination; however, as we noted above, multiple across tissue types. Furthermore, different primer infections (with no evidence of multiple sequence sets may have amplification bias towards particular peaks) may be an alternative explanation. species based on the annealing affinity. These com- These studies, although fairly recent, demonstrate plications may cause the observed Bartonella diver- the potential of this multi-locus approach to charac- sity to differ depending on which marker was used terizing bartonellae without a culturing step. New for amplification, so a single marker may not be a Bartonella species can be described across multiple robust indicator of total Bartonella diversity in a genes showing phylogenetic concordance, or in set of samples (Buffet et al. 2013a). Furthermore, other cases, interesting cases of potential recombin- investigators have observed recombination events ation or multiple infection can be noted. In this even within a single gene (gltA), interfering with way, multi-locus sequencing can be an important phylogenetic inference, but may not be present in first step, focusing on detection and partial genotyp- other sequenced loci (Paziewska et al. 2012;Buffet ing, with other analyses following after to fully char- et al. 2013a). acterize novel or recombinant genotypes by Nevertheless, this approach has one very import- culturing and MLST (or full genome analyses). ant disadvantage – when sequenced loci are in Multi-locus sequencing therefore strikes a valuable conflict regarding the bacterial species identified in balance by providing potentially more robust assess- the sample, one must determine if this is caused by ments of Bartonella diversity than single-locus homologous recombination or the presence of mul- approaches, and is also more accessible to a wider tiple infections. In these cases, researchers may community of researchers than full genomic choose to report the conflicting results as is and approaches since it requires only standard molecular simply note this caveat with the understanding that techniques (PCR and Sanger sequencing).

Metagenome and transcriptome sequencing – the way of ruminants and marine mammals (Table 1). In the the future? 101 studies, bartonellae were cultured from blood, followed by genotyping of the isolates. Of those, cul- Shotgun metagenome and transcriptome sequencing turing work was accompanied with molecular detec- techniques that target many coding loci are becoming tion of Bartonella DNA in tissues only in 16 studies, popular methods for identifying bacteria at the while detection of Bartonella DNA in ectoparasites species level with better phylogenetic resolution at along with culturing bacteria from their hosts was low per-base cost (Venter et al. 2004; Rinke et al. attempted in 21 studies. 2013; Logares et al. 2014;Huget al. 2016) and may represent a new way forward for identifying bacterial pathogens in a large number of samples and tissue Selection of animal tissues for detection and genotyping types. However, the cost of deep sequencing and of Bartonella species absence of comprehensive reference databases (the The most frequent tissue for targeting and genotyp- majority of environmental microorganisms have yet ing Bartonella DNA by PCR and sequencing was to be sequenced) making this approach unavailable blood: 62 studies where only blood was used and for all but the most well-funded laboratories. We three studies where other tissues along with blood look forward to seeing these high-throughput meta- were analysed (Table 1). Other tissues used for genome and transcriptome approaches applied Bartonella genotyping are: spleen (29), liver (8), more frequently (and we expect they will as the cost heart (7), kidney (5), lung (2), and ear, skin and of machinery and computing resources become nail by one study. Besides blood samples, only one more available and affordable), but for now we seek tissue type was analysed in 34 studies, two tissues to make recommendations for genotyping bartonel- in eight studies and more than two tissues in three lae that are accessible to a wider community of studies. researchers. As noted above, a multi-locus sequence Overall, there are limited reports about significant approach may be the best option for many studies variation in detection of Bartonella DNA between and could facilitate broad-scale comparisons of tissues. Razzauti et al. (2015) noted that the choice Bartonella diversity across systems if researchers of organ likely has an important impact on the detec- use a consistent set of markers. tion or misdetection of Bartonella. To explain the huge difference in relative abundance of Bartonella reads detected by 16S MiSeq vs RNA-Seq cited REVIEW OF STUDIES FOCUSING ON GENOTYPING above, the authors used the currently accepted BARTONELLAE FROM ANIMALS SAMPLES model of Bartonella infection described by Harms In order to make recommendations for sequence- and Dehio (2012). This model posits that immedi- based approaches for genotyping bartonellae from ately after infection, bartonellae colonize an archived animal samples, we performed a literature unknown primary niche in the mammalian host, review to identify commonly used genetic markers. most likely vascular endothelial cells. Every 5 days, Based on the results of this survey, we will identify some of the bacteria in the endothelial cells are some candidate markers that could become consist- released into the blood stream, where they infect ery- ent features of the multi-locus sequencing approach throcytes. Then bacteria invade a phagosomal mem- we described above, and thus facilitate valuable brane inside the erythrocytes, where they multiply comparative studies of Bartonella ecology and until they reach a critical population density. At evolution. this point, they simply wait until they are taken up with the erythrocytes by a blood-sucking arthropod. The spleen plays important roles with regard to ery- Analysis of literature on identification of bartonellae in throcytes by removing old erythrocytes, and may animal hosts thereby hold a reserve of erythrocytes that are We surveyed a sample of published literature (>400 highly infected by non-replicating bartonellae, studies) using paired key words ‘bartonella-rodents’, which do not produce RNA molecules. Moreover, ‘bartonella-bats’, ‘bartonella-wildlife’, ‘bartonella- due to its central role in recycling erythrocytes, the cats’, ‘bartonella-dogs’ and ‘bartonella-ectopara- spleen could also store a large amount of degraded sites’. Of the processed literature, 293 studies were DNA of dead bartonellae (Razzauti et al. 2015). selected with available information on application of diverse genetic markers for genotyping of barto- Genetic markers used for identification of bartonellae nellae in identified tissues of vertebrate animals and/or their ectoparasites. These studies report Based on our review, the total number of genetic loci investigations conducted in 79 countries of Africa that have been used for genotyping Bartonella DNA, (19), the USA (11), Asia (21), Australia/Oceania either from bacterial culturing or tissue extracts, (5) and Europe (23), and from a broad diversity of reached 41 (Fig. 1a). The applied markers include animal taxa, including rodents, bats, carnivores, both coding genes and intergenic regions. From

Table 1. Studies aimed to identifying bartonellae in different tissues of vertebrate animals and their ectoparasites (only studies with identified animal hosts are selected; studies with testing off-hosts arthropods are excluded)

DNA extracted directly from tissues Number of DNA extracted from Host studies Culture Blood Heart Spleen Liver Kidney Lung ectoparasites Bats 28 11 9 2 1 0 1 0 11 Rodents 120 61 12 3 23 5 2 2 43 Lagomorphs 4 3 1 0 0 0 0 0 2 Cats 33 4 14 0 0 0 0 2 21 Dogs 15 2 5 0 0 0 0 0 9 Domestic 13 6 4 0 0 0 1 0 7 ruminantsa Wild ruminantsb 15 3 8 2 2 2 0 0 9 Wild carnivoresc 15 6 6 2 2 0 0 0 3 Marine 403000001 mammalsd Other animalse 11 5 3 0 1 0 0 0 3 Total 258 101 65 9 29 2 4 2 109

a Cattle, sheep, goats, horses, camels. b Deer, moose, feral pigs. c Coyotes, foxes, jackals, raccoons, others. d Seals, dolphins, porpoises, sea otters. e Exotic animals, marsupials, primates, birds, turtles.

1994 when the genotyping of Bartonella was Few attempts have been made to culture bartonel- initiated through 2002, the procedure of genotyping lae from arthropod ectoparasites, so identification of was limited to the application of two markers (gltA Bartonella from ectoparasites is typically performed and 16S). The number of genetic loci used has stead- by PCR on extracted DNA. Studies have identified ily increased after 2002 (Fig. 1a), although the Bartonella DNA in a number of ectoparasite majority of studies only use one or two markers groups: fleas (80), ticks (40), lice (13), bat flies (9), (Fig. 1c). Most of the genetic loci were used in deer and sheep keds (9), mites (6), Cimex spp. bugs only a few studies and were not repeated in labora- (2), bees (2) and ants (1). Detection and genotyping tories other than ones where they were proposed. primarily target ITS and gltA, with rpoB being the Only 10 genetic loci were used >10 times in multiple third most common marker [Table 2(c)]. laboratories (gltA,ITS,rpoB, 16S rRNA, ftsZ, groEl, ribC, pap31, nuoG and ssrA) with gltA being Comparison of genetic markers for detection and the most frequently used marker across all studies genotyping of Bartonella DNA in animal tissues surveyed (Fig. 1a). Beyond detection, genetic targets that provide Of 54 publications where identification and genotyp- sufficient sequence diversity to allow differentiation ing of bartonellae in mammalian tissues were con- of Bartonella species are required to fully under- ducted with at least two different genetic markers, stand the distribution and host specificity of only 13 studies provided data for comparing the various Bartonella species and identification of the effectiveness of using different genetic loci strains associated with human illness. The citrate (Table 3). In almost all of these studies, the ITS synthase gene (gltA), originally proposed by target was the most sensitive marker for identifica- Norman et al. (1995), remains the most popular tion of Bartonella DNA in blood. Only one study genetic target for Bartonella detection and is con- focused on detecting and genotyping Bartonella sidered a reliable tool for distinguishing genotypes. DNA in cat blood found the ITS and gltA targets In our review, gltA was used in 48 of 56 of the to be equally productive (Bai et al. 2015b). While studies where one or two markers were applied detecting and genotyping Bartonella in rodent for identification of Bartonella cultures [Table 2 spleens, two studies reported successful identifica- (a)]. Other markers, particularly rpoB and ftsZ, tion in more specimens by targeting the rpoB gene are common when at least four markers are used compared with the gltA (Gundi et al. 2010, 2012b). for direct detection of Bartonella DNA in tissues Birtles and coworkers described the use of PCR- by PCR; ITS is used frequently, comparable with based amplification of ITS fragments to detect and the gltA and more often than rpoB and ftsZ identify bartonellae in the blood of rodents. Direct [Table 2(b)]. detection was of particular use in the longitudinal

Fig. 1. Frequency of genetic loci used for genotyping bartonellae (a), temporal trend in the number of markers used for genotyping (b) and the frequency distribution of the number of markers used across 293 reviewed studies.

survey of Bartonella bacteraemia that involved the Comparison of genetic markers for detection and field collection of very small amounts of blood from genotyping of Bartonella DNA in ectoparasites live, wild rodents (Birtles et al. 2000). As most of Since only few attempts of culturing Bartonella the ITS is non-coding, it is prone to hypervariability, from arthropods have been successful, genotyping and its sequence variation is markedly higher than of Bartonella in insects and acarines relies mostly that observed at other genetic loci (Roux and Raoult, 1995). Although comparison of ITS on detection from extracted DNA by PCR, typically sequences is useful for the allocation of detected relying on only one marker. In spite of a large organisms into one of the recognized Bartonella number of publications reporting investigation of species, detection of a novel ITS sequence can be Bartonella in ectoparasites (>140), we were able to problematic because of the difficulties with sequen- select only 14 publications, which provided data cing of amplified fragments and problems with align- on comparison of at least two genetic markers for ment of the obtained sequences (Knap et al. 2007). genotyping Bartonella in DNA extracted from ITS sequences have many insertions and deletions arthropods (Table 4). In four of the 14 studies, that can complicate phylogenetic analysis. In some ITS was shown to be most sensitive marker for of the studies, screening was conducted by conven- detection and genotyping Bartonella; however, in tional or real-time PCR of ITS, followed by sequen- some other studies, success with gltA gene was cing of additional markers, usually gltA (Miceli et al. similar (De Sousa et al. 2006; Pérez-Martínez et al. 2013; Gutiérrez et al. 2015; Bai et al. 2016). 2009).

Table 2. Summary of markers used for the characterization of bartonellae (a) from cultures, (b) tissues and (c) ectoparasites

Number of markers Number of studies gltA ITS 16S rpoB ftsZ ribC nuoG ssrA Other (a) Cultures 1443725200000 2121153300002 37651400112 49944882010 58856882002 >5 20 19 9 16 19 18 16 5 5 16 Total 100 90 30 35 44 34 20 6 7 22 (b) Tissues 15521233312200 22417143920101 3 17 14 12 3 10 1 1 3 1 1 48861631232 53321322000 >5 3 3 3 2 3 2 2 1 1 2 Total 110 66 60 13 34 11 8 9 5 6 (c) Ectoparasites 17928375212120 23518229532104 310861310211 47571520122 58876881000 >5 2 2 1 1 2 2 2 0 0 1 Total 141 69 80 23 25 17 7 4 3 12

When Morick et al. (2010) genotyped bartonellae and to bacterial species that could inhabit similar eco- in fleas collected from rodents in the Negev Desert logical niches (such as Ehrlichia) (Colborn et al. of Israel using three genetic markers (gltA, ITS and 2010). To identify genus-specific and host-blind rpoB), they found the 313 bp gltA fragment to be primer sets, a whole-genome scan of three the best target for screening fleas for Bartonella and Bartonella genomes (B. henselae, B. quintana and B. for identification to species level. All flea pools that bacilliformis) available at that time was performed were found positive by rpoB or ITS screening were (Colborn et al. 2010), and the NADH dehydrogenase also positive by gltA. Pérez-Martínez et al. (2009) γ subunit (nuoG) primer set was identified and met all investigated 82 fleas collected from cats and dogs in the required conditions. A few years later, another Chile. When rpoB primers were used, Bartonella genetic locus (ssrA), also known as transfer-messen- genotypes were found in four Ctenocephalides felis ger RNA, was proposed as a target for a genus- fleas from cats (4·8%) and in four Pulex irritans fleas specific real-time PCR assay based on analyses on from dogs (4·8%). The same eight samples were posi- whole genomes (Diaz et al. 2012). These markers tive when primers for gltA and ITS were used. None have been used in a number of studies for the detec- of the 82 specimens were positive when primers tar- tion of Bartonella DNA in animal tissues and ectopar- geting the groEL gene were used. Conducting sur- asites, with successful detection at rates similar to veillance of Egyptian fleas for agents of public other loci but still lower than ITS (Gutiérrez et al. health significance, Loftis et al. (2006) detected 2014; Brook et al. 2015; Bai et al. 2015b). more Bartonella-positive fleas using groEL than ITS (17 vs 11) and were successful in conducting Recommendations for marker usage in a multi-locus phylogenetic analysis based on comparison of the sequencing framework groEL sequences rather than ITS sequences. Overall, the Bartonella gltA sequence database in GenBank is the largest and most frequently Contribution of analyses of complete Bartonella updated among the different collections of deposited genomes to primer design sequences, and therefore allows a more accurate Cross-referencing the gltA primer set against the differentiation between Bartonella species and GenBank dataset showed that despite their common strains. A proteomic analysis of gltA indicates that use for Bartonella detection, these primers have most amino substitutions are synonymous, highlight- high cross-reactivity both to potential Bartonella ing the important and critical function of the citrate host DNA (such as Rattus, Mus and Homo sapiens) synthase (gltA) enzyme. Nevertheless, numerous

Table 3. Comparison of genetic loci used for genotyping of Bartonella DNA in animal tissues by their detection frequency

Number of Citation Host/tissue samples ITS gltA rpoB ftsZ ribC nuoG ssrA pap31 groEL

Bai et al. Cat/blood 142 44/142 44/142 29/ (2015b) 142 Gutiérrez Cat/blood 36 23/36 21/23a et al. (2015) Miceli et al. Cat/blood 163 3/163 3/3a 3/3a 0/3a 0/3a (2013) Braga et al. Cat/blood 200 8/200 6/200 3/200 7/ 200 5/200 (2012) 2/200 Kamani et al. Bat/blood 148 76/148b 14/148 0/148 (2014) Brook et al. Bat/blood 76 24/76 5/76 (2015)c Gutiérrez Cattle/blood 50 15/50b 14/ et al. (2014) 50b Carrasco Sea otter/heart 51 23/51 2/51 7/51 et al. (2014) Bai et al. Wild carni- 292 38/292 36/38a 33/ (2016) vores/spleen 38a Ko et al. Deer/spleen 70 20/70 5/70 (2013) Ko et al. Rodents/ 200 124/200 63/200 (2016) spleen Gundi et al. Rodents/ 98 20/98 24/98 (2012b) spleen Gundi et al. Rodents/ 324 42/324 76/324 (2010)c kidney, liver, lung

a The genetic markers used as the second step for genotyping Bartonella in positive DNA after initial screening. b Real-time PCR assay was used for detection of Bartonella DNA without genotyping. c Data are not in publication, but provided from a private communication.

studies have indicated that ITS is a highly sensitive the intervening years, genomes of many Bartonella marker that is invaluable for the detection of species have been sequenced and assembled. Using Bartonella DNA. In order to maximize detection these data, we will evaluate the phylogenetic reso- success and differentiation among related Bartonella lution of a number of candidate loci found in the species, we advocate for a multi-locus sequencing genomes of Bartonella species and compare these approach. Although many markers have been used results to the genetic markers frequently used to in different studies, there is a growing consensus of genotype bartonellae. frequently used markers – specifically, gltA,ITS, rpoB, ftsZ, ribC, groEL, nuoG and ssrA – that are gen- Detection of gene clusters in Bartonella genomes erally capable of differentiating among Bartonella species, particularly when used together in a multi- Genomes from 22 publically available Bartonella locus genotyping framework (La Scola et al. 2003). species were downloaded from GenBank. Every Usage of these markers consistently across studies pair of gene sequences from each genome was will facilitate ecological analyses of Bartonella preva- aligned using the Needleman–Wunsch global align- lence and diversity across systems and comprehensive ment algorithm to all other genes. The resulting phylogenies of known Bartonella species. alignment scores were placed in a square similarity matrix and genes were assigned to clusters using a single-linkage (non-centroid-based, non-greedy), EVALUATION OF PHYLOGENETIC RESOLUTION exhaustive clustering algorithm. The clustering AMONG CANDIDATE LOCI BASED ON ANALYSIS OF threshold was chosen in a way so that each gene BARTONELLA GENOMES cluster is expected to contain the same genes origin- La Scola et al. (2003) used seven protein-coding loci ating from different species. The constituent to genotype Bartonella strains, but as we noted sequences of each gene cluster were then partitioned above, these genes varied considerably in their into separate FASTA files for subsequent analyses power to discriminate among Bartonella species. In (L. Albayrak and C. McKee, unpublished data).

Table 4. Comparison of genetic loci used for genotyping of Bartonella DNA in animal ectoparasites by their detection frequency

Number of Citation Host/ectoparasite samples ITS gltA rpoB ftsZ nuoG groEL

Bai et al. (2015b) Cats/fleas 152 31/152 25/152 Gutiérrez et al. (2015) Cats/fleas 90 68/90 54/90 Brook et al. (2015)a Bats/flies 25 14/25 10/25 12/25 Kamani et al. (2014) Bats/flies 24 10/24b 7/24 7/24 Loftis et al. (2006) Rats/fleas 400 11/400 17/400 Morick et al. (2010) Rodents/fleas 245 66/245 94/245 74/245 Pérez-Martínez et al. Cats and dogs/fleas 82 8/82 8/82 8/82 0/82 (2009) Rojas et al. (2015) Cats and dogs/fleas 72 8/72 5/72 De Sousa et al. (2006) Rodents/fleas 56 4/56 4/56 Bonilla et al. (2009) Humans/lice 153 49/153 39/153 Bonilla et al. (2014)a Humans/lice 11 2/11 1/11 0/11 0/11 Morick et al. (2009) Seals/lice 5 1/5 1/5 Kabeya et al. (2010) Mites/rodents 40 29/40 9/40 Billeter et al. (2011) Ticks 10 2/10 3/10

a Data are not in publication but provided from a private communication. b Real-time PCR assay was used for detection of Bartonella DNA without genotyping.

Ranking of gene clusters by sequence diversity C. McKee, unpublished data). The proportion of unique 32-mers is used here to assess the diversity The overall sequence diversity of each gene cluster of the nucleotide sequences in each cluster and is a was estimated as the ratio between the numbers of useful measure for ranking many gene clusters by unique 32-base long subsequences present in all their sequence diversity. However, this measure sequences in the cluster over the total number of does not necessarily reflect phylogenetic differentia- 32-base long subsequences present in all sequences in the cluster. For each gene cluster, all 32-base tion among congeneric taxa. Hence, we then quan- fi long subsequences (32-mers) from each position in ti ed the sequence diversity and phylogenetic the nucleotide sequences in the FASTA file were resolution of each of these 28 gene clusters using collected. The reverse complements of the extracted additional measures. Sequence diversity was subsequences were added to the complete set of 32- assessed based on the proportion of segregating ’ mers. Unique 32-base long subsequences in the set sites, Watterson s estimator of genetic diversity and were identified and their ratio to the total number nucleotide diversity. Phylogenetic resolution was of subsequences (including duplicates) was calcu- measured by calculating Tamura-Nei sequence dis- lated. We refer to this measurement as the propor- tances and storing the minimum, median and tion of unique 32-mers, and it varies between near maximum distances. All calculations for these mea- zero and one. This measure is equal to one if all sures were performed in MEGA (Kumar et al. 32-base long subsequences identified in the clusters 2016). are unique. Across all of the 28 gene clusters, other measures Gene clusters were sorted in descending order of sequence diversity generally followed a declining based on the proportion of unique 32-mers and trend that corresponded to the ranking by propor- assigned numerical ranks accordingly. The top 10 tion of unique 32-mers (Fig. 2a), and all of the (most diverse) and bottom 10 (most conserved) clus- measures were moderately to highly correlated ters containing sequences from each of 22 Bartonella (0·65 < r < 1). However, there was some variation genomes were selected. We also identified seven present in these estimates that was not captured genes commonly used to genotype bartonellae in the proportion of unique 32-mers, particularly (ftsZ, gltA, groEL, gyrB, nuoG, ribC and rpoB) in the proportion of segregating sites. Tamura-Nei from the ranking, corresponding to the following sequence distances similarly declined across the ranks out of 665 total gene clusters: 171 (ribC), 386 ranking of gene clusters, with some noticeable vari- (gyrB), 543 (nuoG), 565 (gltA), 611 (ftsZ), 635 ation in the median and maximum distances (rpoB) and 658 (groEL). This resulted in a list of (Fig. 2b). The nine genetic loci we analysed (16S 26 gene clusters since groEL was part of the rRNA, ITS, ftsZ, gltA, groEL, gyrB, nuoG, ribC bottom 10. We then added 16S rRNA (rank 665/ and rpoB) fell between the top 10 and bottom 10 665) and ITS (rank 652/665) to this ranking separ- based on the proportion of unique 32-mers, with ately for each Bartonella species for which these the exception of groEL and 16S, which had the sequences were available (L. Albayrak and eighth lowest and the lowest rankings, respectively.

Fig. 2. Measures of sequence diversity (a) and phylogenetic distance (b) across 28 genetic loci occurring in all 22 Bartonella genomes currently available. Loci are numerically ranked in descending order along the bottom axis according to the proportion of unique 32-mers. Commonly used markers for genotyping are labelled with their gene names next to their numerical rank.

These two regions also had low sequence diversity ribC and rpoB), the minimum distances ranged by other measures and low Tamura-Nei distances, were 0·012 for groEL to 0·038 for gyrB and 0·045 indicating that they have poor phylogenetic reso- for ITS. The majority of these minimum distances lution. Overall, the top 10 candidate loci do show were between B. melophagi and B. schoenbuchensis, significantly higher measures of sequence diversity two Bartonella species found in ruminants (deer and phylogenetic distance than the 18 other loci and sheep). (Fig. 3); however, the distributions of minimum Our results largely confirm what La Scola et al. Tamura-Nei distance among these groups do (2003) found; however, our rankings of the overlap (Fig. 1e). These minimum distances corres- markers with the ability to distinguish closely pond to the inverse of the maximum sequence simi- related species were somewhat different, and this is larity that La Scola et al. (2003) used to assess partly due to our usage of entire gene sequences for discriminatory power among loci. Generally, these our measurements (La Scola et al. used only partial minimum distances are small among all loci, gene sequences). In both of our analyses, 16S ranging from 0·002 for 16S rRNA to just 0·085 for rRNA displays the lowest ability to discriminate the top-ranked gene cluster, an unnamed membrane among Bartonella species. The other eight genetic protein (Fig. 2b). Among the eight other commonly loci commonly used for genotyping perform much used markers (ITS, ftsZ, gltA, groEL, gyrB, nuoG, better than 16S rRNA, with minimum distances

Fig. 3. Comparisons of the proportion of unique 32-mers (a), proportion of segregating sites (b), Watterson’s estimator (c), nucleotide diversity (d), Tamura-Nei sequence distance (minimum, median and maximum); (e–f) across 28 genetic loci occurring in all 22 Bartonella genomes currently available. Separate boxes in each plot represent the top 10 loci as ranked by the proportion of unique 32-mers (‘top 10’) and the other 18 loci (‘remaining’).

exceeding 1%, which should be sufficient for identi- Therefore, primer design is a very challenging fying two genotypes as distinct in a phylogenetic problem, especially the design of universal primers analysis, particularly when used together in MLST capable of binding to all possible species (L. analyses. Albayrak and C. McKee, unpublished data). There We were able to identify the presence of candidate is likely a tradeoff in between phylogenetic reso- loci that can discriminate among closely related lution of genes and the ability to design universal Bartonella species better than the commonly used primers, so the clustering of the eight genetic loci markers. We suggest that these loci may be useful commonly used for genotyping (ITS, ftsZ, gltA, for the characterization of Bartonella species and groEL, gyrB, nuoG, ribC and rpoB) between the assessment of phylogenetic relationships; however, most diverse genes and the least diverse genes there currently exist no known primers for amplify- (including 16S rRNA) may be a function of this ing these loci by conventional PCR. Bartonella tradeoff. The other potential disadvantage of using species have high nucleotide diversity across their any of the top 10 most diverse loci for genotyping genomes (Fig. 2a) with few highly conserved bartonellae is that these genes have only been regions, especially in highly diverse genes. sequenced for 22 Bartonella species. There are an

enormous number of potentially new Bartonella phylogenetic resolution for the delineation of bacter- species and genotypes that have been characterized ial species and will provide consistency in the usage by only ITS and/or gltA sequences that considerably of genetic loci that can facilitate a global assessments expand our knowledge of Bartonella diversity; thus, of parasitic bacterial diversity. This approach can be there is an advantage to continued usage of these appropriately modified for the detection and charac- markers to facilitate comparative ecological analyses. terization of parasitic bacteria directly from Switching to different markers would inevitably extracted DNA from a range of sample types, with ignore this diversity and would require considerable the caveat that culturing should be attempted if time and effort to restore. feasible since it is the best way to fully characterize a novel bacterial species. In the following section, we will make specific recommendations for the DISCUSSION detection and genotyping Bartonella in collected samples (Box 1), but we recognize that with some The practicality of single-locus barcoding of bacteria modifications to collection of appropriate samples The utility of DNA barcoding for animal species is and the molecular protocols, this approach is likely partly due to special features of the genetic markers generalizable to a variety of bacterial taxa. it targets. The commonly targeted cytochrome c oxidase I (COI) gene for barcoding animals is Recommendations for the genotyping of Bartonella one of the conserved oxidative phosphorylation subspecies units of the mitochondrial genome. Mitochondria are passed solely from the female parent to The first step to successful detection and potential offspring in animals, so individuals are typically isolation of bartonellae is to collect appropriate haploid at all mitochondrial loci and no recombin- tissues and store them properly [Box 1(a)]. ation occurs in the mitochondrial genome. Gutiérrez et al. (2017) recommend the collection of Additionally, there are sufficiently conserved por- whole blood due to the haemotrophic nature of tions of the COI gene that nearly universal PCR these bacteria using appropriate sterility require- primers have been developed (Hajibabaei et al. ments, especially if culturing is to be attempted. If 2006). Since bacterial genomes are haploid, it may animals are sacrificed and organs pulled out, spleen be tempting to simply find any genetic marker that is probably the most valuable organ for Bartonella has better phylogenetic resolution than 16S rRNA detection, although as we have reviewed above, and use it for DNA barcoding of bacteria. There liver, heart, kidney and lung may show evidence of are four primary problems with this approach: (1) infection, and Bartonella species may vary in abun- increasing sequence diversity in a gene diminishes dance across these tissue types within individuals. the ability to design conserved primers that can be All tissue samples should be transported at low tem- utilized across a broad taxonomic diversity of bac- perature and stored at −20 or −80 °C if not processed teria; (2) homologous recombination is widespread immediately. For ectoparasite samples, storage in (Vos and Didelot, 2008) and can obscure phylogen- 70% ethanol at room temperature is convenient and etic inference (Fraser et al. 2007) and estimates of suitable for molecular detection; however, if cultur- species diversity if only one marker is used; (3) bac- ing is planned, then live specimens are preferred and terial genomes are very flexible in gene content, even additional surface sterilization protocols will be for closely related species (Konstantinidis et al. required (Gutiérrez et al. 2017). During the 2006), so a locus may not exist in all species sur- process of collecting and analysing specimens, we veyed; and (4) multiple species may be present in recommend that investigators attempt to identify the sample, but may not be detectable due to vari- all animals as close as possible to the species level ation in abundance or primer amplification bias. by morphological traits [Box 1(a)]. When morpho- Thus, we argue that for bacterial species, and in par- logical identification is not feasible (e.g. when ticular Bartonella, there is probably no perfect ana- cryptic species of rats and ectoparasites are morpho- logue to single-locus DNA barcoding that could be logically undistinguished or when accurate records used for all sample types. do not exist), then DNA barcoding of host samples There are existing methods, particularly MLST (tissues or whole ectoparasites) at mitochondrial (Stackebrandt et al. 2002), which could balance the loci (e.g. COI) can be incorporated into molecular tradeoffs of culturing bias, phylogenetic resolution, analyses (Hajibabaei et al. 2006). These data are homologous recombination and gene conservation valuable for understanding the ecology and evolu- across species. We believe that MLST of house- tion of Bartonella species, particularly for under- keeping genes (i.e. genes under stabilizing selection standing the host range and specificity, vector encoding metabolic functions) remains a powerful potential and evolutionary codivergence of parasites technique that can be used on uncultured bacteria with their hosts and vectors. to detect evidence of mixed infections and/or We recommend that investigators use homogen- homologous recombination, provide sufficient ization techniques appropriate for particular tissue

may be present in the specimens [Box 1(b)]. Some Box 1. Recommendations for consistent specimens may benefit from pre-enrichment in approaches to genotyping bartonellae liquid growth medium before extraction (Maggi et al. 2005; Duncan et al. 2007; Riess et al. 2008; Bai et al. 2010) or extended lysis steps during the (a) Target multiple animal tissues (e.g. blood, extraction process. Gutiérrez et al. (2017) provide spleen, liver and/or heart) and ectoparasites an excellent review of recommended protocols for for detection. Identify vertebrate and homogenization and extraction. In cases where cul- arthropod hosts to the species level where turing might be attempted, we recommend retaining possible using morphological traits and/or samples of homogenate (either used immediately or barcoding of mtDNA to facilitate ecological frozen at −20 °C or below). As we have advocated analyses. above, culturing is vital for the complete description (b) Use homogenization and DNA extraction of bacterial species and should be attempted in all protocols that maximize yield while reducing studies where appropriate samples are available, the presence of PCR inhibitors. Extended especially if sequence data indicate the presence of lysis or pre-enrichment culture steps may be novel Bartonella species or genotypes [Box 1(c)]. needed for some samples. After Culturing can provide information on valuable homogenization of samples but before DNA traits such as in vitro growth rate, bacterial morph- extraction, retain some samples if culturing ology, presence of multiple coinfecting bartonellae, will be attempted. biochemical profiles, etc. Additional genomic (e.g. (c) When possible, attempt to culture isolates, MLST or whole genome sequencing) analyses that especially when sequence data indicate the clarify evolutionary histories can be facilitated, if presence of novel species or genotypes. Bartonella genotypes are isolated. Genotyping should be regarded as just the For direct detection of Bartonella DNA from first step towards the description of extracted DNA, there are several options for Bartonella species, with additional trait and markers that appear to have good sensitivity. – genomic data providing valuable information Conventional PCR targeting the 16S 23S intergenic for species descriptions. spacer region (ITS) or real-time PCR targeting various loci (e.g. ITS, ssrA, rpoB) are amenable for (d) Screen samples by ITS or real-time PCR screening many samples for potential positives (ITS, rpoB or ssrA) or alternatively, 16S [Box 1(d)]. Primers and protocols for these metagenome or transcriptome sequencing (if approaches are published and are reviewed in available) followed by sequencing of multiple Gutiérrez et al. (2017). We caution against reporting house-keeping genes. results from real-time PCR assays or conventional (e) At the very least, sequence gltA to facilitate PCR in the absence of sequencing, since not all comparison with other studies. primers are entirely specific for Bartonella DNA (f) Sequence at least one additional marker to (Maggi and Breitschwerdt, 2005; Colborn et al. confirm the species identity based on gltA. 2010; Diaz et al. 2012) and may amplify host Conflicting identifications may indicate the DNA, leading to false positives. Usage of next-gen- presence of multiple infections or eration sequencing approaches (e.g. 16S metage- recombinant genotypes. More markers nomics, transcriptomics) are also valuable at this provide more robust results, but at least three stage, especially if investigators are interested in is recommended. describing the microbiome of particular tissues or (g) Additional targets can vary in detection investigating a broad range of pathogenic bacterial success, but rpoB, ftsZ, groEL, ribC, nuoG taxa [Box 1(d)]. However, as we reviewed above, and ssrA are popular (in order of frequency the phylogenetic resolution of 16S sequences is used). Nested PCR reactions can increase limited, so additional genomic loci will need to be fi sensitivity of these markers to be more sequenced to con rm the species identity of targeted comparable with ITS results. bacterial taxa. For accurate genotyping, there are a variety of (h) Attempt to identify the phylogenetic lineage markers that have good phylogenetic resolution (Harms and Dehio, 2012) or associated (La Scola et al. 2003) and validated primer sets Bartonella species complex (Kosoy et al. (Gutiérrez et al. 2017). Based on our review of the 2012) based on sequence data for any novel literature, gltA is the most widely used marker and genotypes. has the most extensive database of sequences available on GenBank. Therefore, we advocate for all or ectoparasite specimens, and follow extraction pro- studies to sequence this marker, at least from any tocols that maximize DNA yield and quality while novel genotypes or species, to support comparisons minimizing the presence of PCR inhibitors that of Bartonella diversity across studies [Box 1(e)].

Nested PCR reactions (Bai et al. 2016) that combine evolutionary radiations through codivergence and published primers (Norman et al. 1995; Birtles and speciation within related hosts, providing informa- Raoult, 1996) can increase the sensitivity of this tion about the biological niche of these Bartonella marker. species. Multiple species complexes linked deeper In addition to gltA, we feel it is important to in evolutionary time may form well-supported acquire additional sequence data from other loci to clades or lineages that illuminate the longer term confirm the species identification by gltA [Box 1 diversification processes of this diverse genus. (f)]. Phylogenetic concordance among loci may be Researchers describing new Bartonella species or sufficient to describe novel, candidate Bartonella genotypes could increase the impact of their species (Lilley et al. 2015; Martin-Alonso et al. findings by making these substantial evolutionary 2016), which can be further described after and ecological connections, particularly when clin- culturing isolates. Conflicts among multiple loci ical cases of bartonellosis can be traced to a potential may indicate the presence of multiple infections zoonotic origin. or recombinant infections – molecular cloning or examination of multiple peaks in chromatograms Concluding remarks may be able to distinguish these scenarios. In general, three loci should be sufficient to accurately Using Bartonella bacteria as examples, we have genotype Bartonella if a single infection is present highlighted the substantial challenges that exist in or detect mixed infections; however, more loci may the accurate genotyping of bacteria from environ- produce more robust results or distinguish mental samples. Issues related to isolation of cul- Bartonella genotypes that may be closely related to tures, homologous recombination, coinfections, known species. The number of markers to use will sensitivity and phylogenetic resolution of molecular depend on the phylogenetic resolution needed markers, variation in detection across different (most MLST studies require only 5–9 markers), tissues, and variation in marker usage across and more markers, especially uncommon markers, studies are certainly not restricted to studies of may have limited usefulness if not repeated in Bartonella. For environmental samples stored in other laboratories. Additional targets can vary in laboratories and museums around the world, our sensitivity and amplification bias for particular recommendations for sensitive detection assays species [Box 1(g)], but rpoB, ftsZ, groEL, ribC, (including real-time PCR or high-throughput meta- nuoG and ssrA are popular (in order of frequency genomics), followed by conventional PCR amplifica- used based on our literature review) and have pub- tion and sequencing of multiple house-keeping lished primers and amplification protocols genes are surely applicable to a wide array of other (Gutiérrez et al. 2017), including some nested pro- zoonotic bacteria. These may include many proteo- tocols (Zeaiter et al. 2002; Colborn et al. 2010; Bai bacteria (e.g. Anaplasma, Bordatella, Brucella, et al. 2016). Our analysis of 22 Bartonella genomes Burkholderia, Campylobacter, Coxiella, Ehrlichia, indicates that there may be single loci that are Yersinia, Francisella, Helicobacter, Legionella, capable of distinguishing among Bartonella species Neorickettsia, Orientia, Pasteurella, Pseudomonas, better than these popular markers alone, but Rickettsia and Wolbachia), spirochetes (e.g. primer design for these regions will be challenging Borrelia, Leptospira, Treponema) and other bacteria due to their sequence diversity (i.e. with few con- (e.g. Chlamydia, Listeria, Mycobacterium, served regions), and the utility of these sequences Mycoplasma). Although recommendations for the will lag behind these popular markers unless many collection of appropriate animal samples and laboratories adopt them. Researchers should experi- molecular markers used for detection and character- ment with multiple markers and modified protocols ization will vary according to each bacteria (and to find the best ones for their sample type, but some likely requires standardization as we observed with consistency among sequenced markers across Bartonella), the general process of direct detection studies will facilitate comparative studies of from extracted DNA, multi-locus sequencing and Bartonella prevalence and diversity across systems. subsequent attempts to culture (followed by add- Finally, our understanding of Bartonella ecology itional genomic or biochemical characterization) is and evolution would benefit from the increased generalizable. Databases already exist, e.g. http:// description of Bartonella phylogenetic lineages www.mlst.net/databases/default.asp and https:// (Harms and Dehio, 2012) and species complexes pubmlst.org/databases/), that contain primers and (Kosoy et al. 2012), especially for novel genotypes protocols for multi-locus sequencing approaches [Box 1(h)]. Species complexes can include clusters for many of the zoonotic bacteria listed above. genetically similar species found in a group of Standardized genotyping approaches will greatly related hosts, such as B. elizabethae, Bartonella expand our knowledge of the phylogenetic diversity queenslandensis, Bartonella rattimassiliensis and B. and ecology of parasitic bacteria infecting animals, tribocorum associated with murine rodents. These and help to measure and mitigate the risks posed species complexes can indicate the presence of by these bacteria to public health.

ACKNOWLEDGMENTS Bonilla, D. L., Kabeya, H., Henn, J., Kramer, V. L. and Kosoy, M. Y. (2009). Bartonella quintana in body lice and head lice from homeless The findings and conclusions in this report are those of the persons, San Francisco, California, USA. Emerging Infectious Diseases 15, authors and do not necessarily represent the official pos- 912–915. ition of CDC. Bonilla, D. L., Cole-Porse, C., Kjemtrup, A., Osikowicz, L. and Kosoy, M. (2014). Risk factors for human lice and bartonellosis among the homeless, San Francisco, California, USA. Emerging Infectious FINANCIAL SUPPORT Diseases 20, 1645–1651. Braga, M. do S. C. de O., Diniz, P. P. V. de P., André, M. R., de Numerous investigations of zoonotic bacterial agents Bortoli, C. P. and Machado, R. Z. (2012). Molecular characterisation of around the globe were supported by the CDC’ Global Bartonella species in cats from São Luís, state of Maranhão, north- Disease Detection program. eastern Brazil. Memórias do Instituto Oswaldo Cruz 107, 772–777. Brook, C. E., Bai, Y., Dobson, A. P., Osikowicz, L. M., Ranaivoson, H. C., Zhu, Q., Kosoy, M. Y. and Dittmar, K. (2015). Bartonella spp. in REFERENCES fruit bats and blood-feeding ectoparasites in Madagascar. PLoS Neglected Tropical Diseases 9, e0003532. André, A., Mouton, A., Millien, V. and Michaux, J. (2017). Liver micro- Brook, C. E., Bai, Y., Yu, E. O., Ranaivoson, H. C., Shin, H., biome of Peromyscus leucopus, a key reservoir host species for emerging Dobson, A. P., Metcalf, C. J. E., Kosoy, M. Y. and Dittmar, K. infectious diseases in North America. Infection, Genetics and Evolution 52, (2017). Elucidating transmission dynamics and host-parasite-vector rela- 10–18. tionships for rodent-borne Bartonella spp. in Madagascar. Epidemics In Auch, A. F., von Jan, M., Klenk, H.-P. and Göker, M. (2010). Digital press. DNA–DNA hybridization for microbial species delineation by means of Buffet, J.-P., Pisanu, B., Brisse, S., Roussel, S., Félix, B., Halos, L., genome-to-genome sequence comparison. Standards in Genomic Sciences Chapuis, J.-L. and Vayssier-Taussat, M. (2013a). Deciphering 2, 117–134. Bartonella diversity, recombination, and host specificity in a rodent com- Bai, Y., Kosoy, M. Y., Calisher, C. H., Cully, J. F. and Collinge, S. K. munity. PLoS ONE 8, e68956. (2009). Effects of rodent community diversity and composition on preva- Buffet, J.-P., Kosoy, M. Y. and Vayssier-Taussat, M. (2013b). Natural lence of an endemic bacterial pathogen – Bartonella. Biodiversity 10,3–11. history of Bartonella-infecting rodents in light of new knowledge on gen- Bai, Y., Kosoy, M. Y., Boonmar, S., Sawatwong, P., omics, diversity and evolution. Future Microbiology 8, 1117–1128. Sangmaneedet, S. and Peruski, L. F. (2010). Enrichment culture and Carrasco, S. E., Chomel, B. B., Gill, V. A., Kasten, R. W., Maggi, R. molecular identification of diverse Bartonella species in stray dogs. G., Breitschwerdt, E. B., Byrne, B. A., Burek-Huntington, K. A., Veterinary Microbiology 146, 314–319. Miller, M. A., Goldstein, T. and Mazet, J. A. K. (2014). Novel Bai, Y., Kosoy, M. Y., Diaz, M. H., Winchell, J., Baggett, H., Bartonella infection in northern and southern sea otters (Enhydra lutris Maloney, S. A., Boonmar, S., Bhengsri, S., Sawatwong, P. and kenyoni and Enhydra lutris nereis). Veterinary Microbiology 170, 325–334. Peruski, L. F. (2012). Bartonella vinsonii subsp. arupensis in humans, Chaloner, G. L., Palmira Ventosilla and Birtles, R. J. (2011). Multi-locus Thailand. Emerging Infectious Diseases 18, 989–991. sequence analysis reveals profound genetic diversity among isolates of the Bai, Y., Hayman, D. T. S., McKee, C. D. and Kosoy, M. Y. (2015a). human pathogen Bartonella bacilliformis. PLoS Neglected Tropical Classification of Bartonella strains associated with straw-colored fruit Diseases 5, e1248. bats (Eidolon helvum) across Africa using a multi-locus sequence typing Colborn, J. M., Kosoy, M. Y., Motin, V. L., Telepnev, M. V., platform. PLoS Neglected Tropical Diseases 9, e0003478. Valbuena, G., Myint, K. S. A., Fofanov, Y., Putonti, C., Feng, C. and Bai, Y., Rizzo, M. F., Alvarez, D., Moran, D., Peruski, L. F. and Peruski, L. F. (2010). Improved detection of Bartonella DNA in mamma- Kosoy, M. (2015b). Coexistence of Bartonella henselae and B. clarridgeiae lian hosts and arthropod vectors by real-time PCR using the NADH in populations of cats and their fleas in Guatemala. Journal of Vector dehydrogenase gamma subunit (nuoG). Journal of Clinical Microbiology Ecology 40, 327–332. 48, 4630–4633. Bai, Y., Gilbert, A., Fox, K., Osikowicz, L. and Kosoy, M. (2016). Curtis, T. P., Sloan, W. T. and Scannell, J. W. (2002). Estimating pro- Bartonella rochalimae and B. vinsonii subsp. berkhoffii in wild carnivores karyotic diversity and its limits. Proceedings of the National Academy of from Colorado, USA. Journal of Wildlife Diseases 52, 844–849. Sciences of the United States of America 99, 10494–10499. Banskar, S., Bhute, S. S., Suryavanshi, M. V., Punekar, S. and DeLong, E. F. and Pace, N. R. (2001). Environmental diversity of Shouche, Y. S. (2016). Microbiome analysis reveals the abundance of bac- Bacteria and Archaea. Systematic Biology 50, 470–478. terial pathogens in Rousettus leschenaultii guano. Scientific Reports 6, 36948. De Sousa, R., Edouard-Fournier, P., Santos-Silva, M., Amaro, F., Bellemain, E., Carlsen, T., Brochmann, C., Coissac, E., Taberlet, P. Bacellar, F. and Raoult, D. (2006). Molecular detection of Rickettsia and Kauserud, H. (2010). ITS as an environmental DNA barcode for felis, Rickettsia typhi and two genotypes closely related to Bartonella eliza- fungi: an in silico approach reveals potential PCR biases. BMC bethae. American Journal of Tropical Medicine and Hygiene 75, 727–731. Microbiology 10, 189. Diaz, M. H., Bai, Y., Malania, L., Winchell, J. M. and Kosoy, M. Y. Berglund, E. C., Ellegaard, K., Granberg, F., Xie, Z., Maruyama, S., (2012). Development of a novel genus-specific real-time PCR assay for Kosoy, M. Y., Birtles, R. J. and Andersson, S. G. E. (2010). Rapid diver- detection and differentiation of Bartonella species and genotypes. Journal sification by recombination in Bartonella grahamii from wild rodents in of Clinical Microbiology 50, 1645–1649. Asia contrasts with low levels of genomic divergence in Northern Europe Dietrich, M., Kearney, T., Seamark, E. C. J. and Markotter, W. and America. Molecular Ecology 19, 2241–2255. (2017). The excreted microbiota of bats: evidence of niche specialisation Billeter, S. A., Levy, M. G., Chomel, B. B. and Breitschwerdt, E. B. based on multiple body habitats. FEMS Microbiology Letters 364, fnw284. (2008). Vector transmission of Bartonella species with emphasis on the Duncan, A. W., Maggi, R. G. and Breitschwerdt, E. B. (2007). A com- potential for tick transmission. Medical and Veterinary Entomology 22,1–15. bined approach for the enhanced detection and isolation of Bartonella Billeter, S. A., Cáceres, A. G., Gonzales-Hidalgo, J., Luna-Caypo, D. species in dog blood samples: pre-enrichment liquid culture followed by and Kosoy, M. Y. (2011). Molecular detection of Bartonella species in PCR and subculture onto agar plates. Journal of Microbiological Methods ticks from Peru. Journal of Medical Entomology 48, 1257–1260. 69, 273–281. Birtles, R. J. and Raoult, D. (1996). Comparison of partial citrate syn- Dykhuizen, D. (2005). Species numbers in bacteria. Proceedings of the thase gene (gltA) sequences for phylogenetic analysis of Bartonella California Academy of Sciences 56,62–71. species. International Journal of Systematic Bacteriology 46, 891–897. Fraser, C., Hanage, W. P. and Spratt, B. G. (2007). Recombination and Birtles, R. J., Hazel, S., Bown, K., Raoult, D., Begon, M. and the nature of bacterial speciation. Science 315, 476–480. Bennett, M. (2000). Subtyping of uncultured bartonellae using sequence Galan, M., Razzauti, M., Bard, E., Bernard, M., Brouat, C., comparison of 16 S/23 S rRNA intergenic spacer regions amplified directly Charbonnel, N., Dehne-Garcia, A., Loiseau, A., Tatard, C., from infected blood. Molecular and Cellular Probes 14,79–87. Tamisier, L., Vayssier-Taussat, M., Vignes, H. and Cosson, J.-F. Bittar, F., Keita, M. B., Lagier, J.-C., Peeters, M., Delaporte, E. and (2016). 16S rRNA amplicon sequencing for epidemiological surveys of Raoult, D. (2014). Gorilla gorilla gorilla gut: a potential reservoir of patho- bacteria in wildlife. mSystems 1, e00032–16. genic bacteria as revealed using culturomics and molecular tools. Scientific Goris, J., Konstantinidis, K. T., Klappenbach, J. A., Coenye, T., Reports 4, 7174. Vandamme, P. and Tiedje, J. M. (2007). DNA–DNA hybridization Blaxter, M. (2016). Imagining Sisyphus happy: DNA barcoding and the values and their relationship to whole-genome sequence similarities. unnamed majority. Philosophical Transactions of the Royal Society B: International Journal of Systematic and Evolutionary Microbiology 57, Biological Sciences 371, 20150329. 81–91.

Gundi, V. A. K. B., Kosoy, M. Y., Myint, K. S., Shrestha, S. K., characterization of Bartonella species in Republic of Korea. Shrestha, M. P., Pavlin, J. A. and Gibbons, R. V. (2010). Prevalence Transboundary and Emerging Diseases 63,56–67. and genetic diversity of Bartonella species detected in different tissues of Konstantinidis, K. T. and Tiedje, J. M. (2005). Genomic insights small mammals in Nepal. Applied and Environmental Microbiology 76, that advance the species definition for prokaryotes. Proceedings of the 8247–8254. National Academy of Sciences of the United States of America 102, Gundi, V. A. K. B., Billeter, S. A., Rood, M. P. and Kosoy, M. Y. 2567–2572. (2012a). Bartonella spp. in rats and zoonoses, Los Angeles, California, Konstantinidis, K. T., Ramette, A. and Tiedje, J. M. (2006). The USA. Emerging Infectious Diseases 18, 631–633. bacterial species definition in the genomic era. Philosophical Transactions Gundi, V. A. K. B., Kosoy, M. Y., Makundi, R. H. and Laudisoit, A. of the Royal Society of London. Series B: Biological Sciences 361, 1929–1940. (2012b). Identification of diverse Bartonella genotypes among small Koskela, K. A., Kalin-Mänttäri, L., Hemmilä, H., Smura, T., mammals from Democratic Republic of Congo and Tanzania. American Kinnunen, P. M., Niemimaa, J., Henttonen, H. and Nikkari, S. Journal of Tropical Medicine and Hygiene 87, 319–326. (2017). Metagenomic evaluation of bacteria from voles. Vector-Borne and Gutiérrez, R., Cohen, L., Morick, D., Mumcuoglu, K. Y., Harrus, S. Zoonotic Diseases 17, 123–133. and Gottlieb, Y. (2014). Identification of different Bartonella species in Kosoy, M. Y., Murray, M., Gilmore, R. D., Jr., Bai, Y. and Gage, K. L. the cattle tail louse (Haematopinus quadripertusus) and in cattle blood. (2003). Bartonella strains from ground squirrels are identical to Bartonella Applied and Environmental Microbiology 80, 5477–5483. washoensis isolated from a human patient. Journal of Clinical Microbiology Gutiérrez, R., Nachum-Biala, Y. and Harrus, S. (2015). Relationship 41, 645–650. between the presence of Bartonella species and bacterial loads in cats and Kosoy, M. Y., Hayman, D. T. S. and Chan, K.-S. (2012). Bartonella bac- cat fleas (Ctenocephalides felis) under natural conditions. Applied and teria in nature: where does population variability end and a species start? Environmental Microbiology 81, 5613–5621. Infection, Genetics and Evolution 12, 894–904. Gutiérrez, R., Vayssier-Taussat, M., Buffet, J.-P. and Harrus, S. Kosoy, M., Bai, Y., Enscore, R., Rizzo, M. R., Bender, S., Popov, V., (2017). Guidelines for the isolation, molecular detection, and characteriza- Albayrak, L., Fofanov, Y. and Chomel, B. (2016). Bartonella melophagi tion of Bartonella species. Vector-Borne and Zoonotic Diseases 17,42–50. in blood of domestic sheep (Ovis aries) and sheep keds (Melophagus ovinus) Guy, L., Nystedt, B., Sun, Y., Näslund, K., Berglund, E. C. and from the southwestern US: cultures, genetic characterization, and eco- Andersson, S. G. E. (2012). A genome-wide study of recombination rate logical connections. Veterinary Microbiology 190,43–49. variation in Bartonella henselae. BMC Evolutionary Biology 12, 65. Kumar, S., Stecher, G. and Tamura, K. (2016). MEGA7: molecular Hajibabaei, M., Janzen, D. H., Burns, J. M., Hallwachs, W. and evolutionary genetics analysis version 7.0 for bigger datasets. Molecular Hebert, P. D. N. (2006). DNA barcodes distinguish species of tropical Biology and Evolution 33, 1870–1874. Lepidoptera. Proceedings of the National Academy of Sciences of the Kyrpides, N. C., Hugenholtz, P., Eisen, J. A., Woyke, T., Göker, M., United States of America 103, 968–971. Parker, C. T., Amann, R., Beck, B. J., Chain, P. S. G., Chun, J., Hanage, W. P., Spratt, B. G., Turner, K. M. and Fraser, C. (2006). Colwell, R. R., Danchin, A., Dawyndt, P., Dedeurwaerdere, T., Modelling bacterial speciation. Philosophical Transactions of the Royal DeLong, E. F., Detter, J. C., De Vos, P., Donohue, T. J., Dong, X.-Z., Society of London. Series B, Biological Sciences 361, 2039–2044. Ehrlich, D. S., Fraser, C., Gibbs, R., Gilbert, J., Gilna, P., Harms, A. and Dehio, C. (2012). Intruders below the radar: molecular Glöckner, F. O., Jansson, J. K., Keasling, J. D., Knight, R., pathogenesis of Bartonella spp. Clinical Microbiology Reviews 25,42–78. Labeda, D., Lapidus, A., et al. (2014). Genomic encyclopedia of bacteria Harrus, S., Bar-Gal, G. K., Golan, A., Elazari-Volcani, R., Kosoy, M. and archaea: sequencing a myriad of type strains. PLoS Biology 12, Y., Morick, D., Avidor, B. and Baneth, G. (2009). Isolation and genetic e1001920. characterization of a Bartonella strain closely related to Bartonella tribo- La Scola, B., Zeaiter, Z., Khamis, A. and Raoult, D. (2003). Gene- corum and Bartonella elizabethae in Israeli commensal rats. American sequence-based criteria for species definition in bacteriology: the Journal of Tropical Medicine and Hygiene 81,55–58. Bartonella paradigm. Trends in Microbiology 11, 318–321. Hebert, P. D. N., Cywinska, A., Ball, S. L. and DeWaard, J. R. (2003). Lebonah, D. E., Dileep, A., Chandrasekhar, K., Sreevani, S., Biological identifications through DNA barcodes. Proceedings of the Royal Sreedevi, B. and Pramoda Kumari, J. (2014). DNA barcoding on bac- Society of London. Series B: Biological Sciences 270, 313–321. teria: a review. Advances in Biology 2014, 541787. Hug, L. A., Baker, B. J., Anantharaman, K., Brown, C. T., Probst, A. Lei, B. R. and Olival, K. J. (2014). Contrasting patterns in mammal–bac- J., Castelle, C. J., Butterfield, C. N., Hernsdorf, A. W., Amano, Y., teria coevolution: Bartonella and Leptospira in bats and rodents. PLoS Ise, K., Suzuki, Y., Dudek, N., Relman, D. A., Finstad, K. M., Neglected Tropical Diseases 8, e2738. Amundson, R., Thomas, B. C. and Banfield, J. F. (2016). A new view Lilley, T. M., Veikkolainen, V. and Pulliainen, A. T. (2015). Molecular of the tree of life. Nature Microbiology 1,1–6. detection of Candidatus Bartonella hemsundetiensis in bats. Vector-Borne Kabeya, H., Colborn, J. M., Bai, Y., Lerdthusnee, K., Richardson, J. and Zoonotic Diseases 15, 706–708. H., Maruyama, S. and Kosoy, M. Y. (2010). Detection of Bartonella Locey, K. J. and Lennon, J. T. (2016). Scaling laws predict global micro- tamiae DNA in ectoparasites from rodents in Thailand and their sequence bial diversity. Proceedings of the National Academy of Sciences of the United similarity with bacterial cultures from Thai patients. Vector Borne and States of America 113, 5970–5975. Zoonotic Diseases (Larchmont, N.Y.) 10, 429–434. Loftis, A. D., Reeves, W. K., Szumlas, D. E., Abbassy, M. M., Kamani, J., Baneth, G., Mitchell, M., Mumcuoglu, K. Y., Helmy, I. M., Moriarity, J. R. and Dasch, G. A. (2006). Surveillance Gutiérrez, R. and Harrus, S. (2014). Bartonella species in bats of Egyptian fleas for agents of public health significance: Anaplasma, (Chiroptera) and bat flies (Nycteribiidae) from Nigeria, West Africa. Bartonella, Coxiella, Ehrlichia, Rickettsia, and Yersinia pestis. American Vector-Borne and Zoonotic Diseases 14, 625–632. Journal of Tropical Medicine and Hygiene 75,41–48. Kandelaki, G., Malania, L., Bai, Y., Chakvetadze, N., Katsitadze, G., Logares, R., Sunagawa, S., Salazar, G., Cornejo-Castillo, F. M., Imnadze, P., Nelson, C., Harrus, S. and Kosoy, M. (2016). Human Ferrera, I., Sarmento, H., Hingamp, P., Ogata, H., de Vargas, C., lymphadenopathy caused by ratborne Bartonella, Tbilisi, Georgia. Lima-Mendez, G., Raes, J., Poulain, J., Jaillon, O., Wincker, P., Emerging Infectious Diseases 22, 544–546. Kandels-Lewis, S., Karsenti, E., Bork, P. and Acinas, S. G. (2014). Keim, P., Pearson, T. and Okinaka, R. (2008). Microbial forensics: Metagenomic 16S rDNA Illumina tags are a powerful alternative to ampli- DNA fingerprinting of Bacillus anthracis (Anthrax). Analytical Chemistry con sequencing to explore diversity and structure of microbial communi- 80, 4791–4800. ties. Environmental Microbiology 16, 2659–2671. Klangthong, K., Promsthaporn, S., Leepitakrat, S., Schuster, A. L., Maggi, R. G. and Breitschwerdt, E. B. (2005). Potential limitations of McCardle, P. W., Kosoy, M. and Takhampunya, R. (2015). The distri- the 16S-23S rRNA intergenic region for molecular detection of bution and diversity of Bartonella species in rodents and their ectoparasites Bartonella species. Journal of Clinical Microbiology 43, 1171–1176. across Thailand. PLoS ONE 10, e0140856. Maggi, R. G., Duncan, A. W. and Breitschwerdt, E. B. (2005). Novel Knap, N., Duh, D., Birtles, R., Trilar, T., Petrovec, M. and Avšič- chemically modified liquid medium that will support the growth of seven Županc, T. (2007). Molecular detection of Bartonella species infecting Bartonella species. Journal of Clinical Microbiology 43, 2651–2655. rodents in Slovenia. FEMS Immunology & Medical Microbiology 50, Manichanh, C., Rigottier-Gois, L., Bonnaud, E., Gloux, K., 45–50. Pelletier, E., Frangeul, L., Nalin, R., Jarrin, C., Chardon, P., Ko, S., Kim, S., Kang, J., Won, S., Lee, H., Shin, N., Choi, K., Marteau, P., Roca, J. and Dore, J. (2006). Reduced diversity of faecal Youn, H. and Chae, J. (2013). Molecular detection of Bartonella grahamii microbiota in Crohn’s disease revealed by a metagenomic approach. Gut and B. schoenbuchensis-related species in Korean water deer (Hydropotes 55, 205–211. inermis argyropus). Vector-Borne and Zoonotic Diseases 13, 415–418. Martin-Alonso, A., Houemenou, G., Abreu-Yanes, E., Valladares, B., Ko, S., Kang, J.-G., Kim, H.-C., Klein, T. A., Choi, K.-S., Song, J.-W., Feliu, C. and Foronda, P. (2016). Bartonella spp. in small mammals, Youn, H.-Y. and Chae, J. S. (2016). Prevalence, isolation and molecular Benin. Vector-Borne and Zoonotic Diseases 16, 229–237.

Meier-Kolthoff, J. P., Hahnke, R. L., Petersen, J., Scheuner, C., Vesbach, C., Hedlund, B. P., Guy, L. and Ettema, T. J. (2015). Michael, V., Fiebig, A., Rohde, C., Rohde, M., Fartmann, B., Exploring microbial dark matter to resolve the deep archaeal ancestry of Goodwin, L. A., Chertkov, O., Reddy, T. B. K., Pati, A., Ivanova, N. eukaryotes. Philosophical Transactions of the Royal Society of London. N., Markowitz, V., Kyrpides, N. C., Woyke, T., Göker, M. and Series B, Biological Sciences 370, 20140328. Klenk, H.-P. (2014a). Complete genome sequence of DSM 30083T, the Schloss, P. D. and Handelsman, J. (2004). Status of the microbial type strain (U5/41T) of Escherichia coli, and a proposal for delineating sub- census. Microbiology and Molecular Biology Reviews 68, 686–691. species in microbial taxonomy. Standards in Genomic Sciences 9,2. Schoch, C. L., Seifert, K. A., Huhndorf, S., Robert, V., Spouge, J. L., Meier-Kolthoff, J. P., Klenk, H.-P. and Göker, M. (2014b). Taxonomic Levesque, C. A., Chen, W. and Consortium, F. B. (2012). Nuclear ribo- use of DNA G+C content and DNA–DNA hybridization in the genomic somal internal transcribed spacer (ITS) region as a universal DNA barcode age. International Journal of Systematic and Evolutionary Microbiology 64, marker for Fungi. Proceedings of the National Academy of Sciences of the 352–356. United States of America 109, 6241–6246. Miceli, N. G., Gavioli, F. A., Goncalves, L. R., Andre, M. R., Segers, F. H., Kešnerová, L., Kosoy, M. and Engel, P. (2017). Genomic Sousa, V. R. F., de Sousa, K. C. M. and Machado, R. Z. (2013). changes associated with the evolutionary transition of an insect gut symbiont Molecular detection of feline arthropod-borne pathogens in cats in into a blood-borne pathogen. The ISME Journal 11, 1232–1244. Cuiaba, state of Mato Grosso, central-western region of Brazil. Revista Stackebrandt, E., Frederiksen, W., Garrity, G. M., Grimont, P. A. D., Brasileira de Parasitologia Veterinária 22, 385–390. Kämpfer, P., Maidem, M. C. J., Nesme, X., Rosselló-Morá, Miller, C. S., Baker, B. J., Thomas, B. C., Singer, S. W. and Swings, J., Trüper, H. G., Vauterin, L., Ward, A. C. and Banfield, J. F. (2011). EMIRGE: reconstruction of full-length ribosomal Whitman, W. B. (2002). Report of the ad hoc committee for the re-evalu- genes from microbial community short read sequencing data. Genome ation of the species definition in bacteriology. International Journal of Biology 12, R44. Systematic and Evolutionary Microbiology 52, 1043–1047. Morick, D., Osinga, N., Gruys, E. and Harrus, S. (2009). Identification Tindall, B. J., Rosselló-Móra, R., Busse, H.-J., Ludwig, W. and of a Bartonella species in the harbor seal (Phoca vitulina) and in seal lice Kämpfer, P. (2010). Notes on the characterization of prokaryote strains (Echinophtirius horridus). Vector-Borne and Zoonotic Diseases 9, 751–753. for taxonomic purposes. International Journal of Systematic and Morick, D., Krasnov, B. R., Khokhlova, I. S., Shenbrot, G. I., Evolutionary Microbiology 60, 249–266. Kosoy, M. Y. and Harrus, S. (2010). Bartonella genotypes in fleas Tsai, Y.-L., Chang, C.-C., Chuang, S.-T. and Chomel, B. B. (2011). (Insecta: Siphonaptera) collected from rodents in the Negev Desert, Bartonella species and their ectoparasites: selective host adaptation or Israel. Applied and Environmental Microbiology 76, 6864–6869. strain selection between the vector and the mammalian host? Norman, A. F., Regnery, R., Jameson, P., Greene, C. and Krause, D. Comparative Immunology, Microbiology and Infectious Diseases 34, 299–314. C. (1995). Differentiation of Bartonella-like isolates at the species level by Urushadze, L., Bai, Y., Osikowicz, L. M., McKee, C. D., PCR-restriction fragment length polymorphism in the citrate synthase Kandaurov, A., Kuzmin, I. V., Sidamonidze, K., Putkaradze, D., gene. Journal of Clinical Microbiology 33, 1797–1803. Imnadze, P. and Kosoy, M. Y. (2017). Prevalence, diversity, and host Osikowicz, L. M., Billeter, S. A., Rizzo, M. F., Rood, M. P., associations of Bartonella strains in bats from Georgia (Caucasus). PLoS Freeman, A. N., Burns, J. E., Hu, R., Juieng, P., Loparev, V. and Neglected Tropical Diseases 11, e0005428. Kosoy, M. (2016). Distribution and diversity of Bartonella washoensis Veikkolainen, V., Vesterinen, E. J., Lilley, T. M. and Pulliainen, A. strains in ground squirrels from California and their potential link to T. (2014). Bats as reservoir hosts of human bacterial pathogen, Bartonella human cases. Vector-Borne and Zoonotic Diseases 16, 683–690. mayotimonensis. Emerging Infectious Diseases 20, 960–967. Paziewska, A., Harris, P. D., Zwolinska,́ L., Bajer, A. and Sinski,́ E. Venter, J. C., Remington, K., Heidelberg, J. F., Halpern, A. L., (2011). Recombination within and between species of the alpha proteobac- Rusch, D., Eisen, J. A., Wu, D., Paulsen, I., Nelson, K. E., terium Bartonella infecting rodents. Microbial Ecology 61, 134–145. Nelson, W., Fouts, D. E., Levy, S., Knap, A. H., Lomas, M. W., Paziewska, A., Sinski,́ E. and Harris, P. D. (2012). Recombination, Nealson, K., White, O., Peterson, J., Hoffman, J., Parsons, R., diversity and allele sharing of infectivity proteins between Bartonella Baden-Tillson, H., Pfannkoch, C., Rogers, Y.-H. and Smith, H. O. species from rodents. Microbial Ecology 64, 525–536. (2004). Environmental genome shotgun sequencing of the Sargasso Sea. Pérez-Martínez, L., Venzal, J. M., González-Acuña, D., Portillo, A., Science 304,66–74. Blanco, J. R. and Oteo, J. A. (2009). Bartonella rochalimae and other Vos, M. (2009). Why do bacteria engage in homologous recombination? Bartonella spp. in Fleas, Chile. Emerging Infectious Diseases 15, 1150–1152. Trends in Microbiology 17, 226–232. Qiu, Y., Nakao, R., Ohnuma, A., Kawamori, F. and Sugimoto, C. Vos, M. and Didelot, X. (2008). A comparison of homologous recombin- (2014). Microbial population analysis of the salivary glands of ticks; a pos- ation rates in bacteria and archaea. The ISME Journal 3, 199–208. sible strategy for the surveillance of bacterial pathogens. PLoS ONE 9, Wayne, L. G., Brenner, D. J., Colwell, R. R., Grimont, P. A. D., e103961. Kandler, O., Krichevsky, M. I., Moore, L. H., Moore, W. E. C., Razzauti, M., Galan, M., Bernard, M., Maman, S., Klopp, C., Murray, R. G. E., Stackebrandt, E., Starr, M. P. and Trüper, H. G. Charbonnel, N., Vayssier-Taussat, M., Eloit, M. and Cosson, J.-F. (1987). Report of the ad hoc committee on reconciliation of approaches (2015). A comparison between transcriptome sequencing and 16S metage- to bacterial systematics. International Journal of Systematic and nomics for detection of bacterial pathogens in wildlife. PLoS Neglected Evolutionary Microbiology 37, 463–464. Tropical Diseases 9, e0003929. Welch, D. F., Carroll, K. C., Hofmeister, E. K., Persing, D. H., Richter, M. and Rosselló-Móra, R. (2009). Shifting the genomic gold Robison, D. A., Steigerwalt, A. G. and Brenner, D. J. (1999). standard for the prokaryotic species definition. Proceedings of the Isolation of a new subspecies, Bartonella vinsonii subsp. arupensis, from a National Academy of Sciences of the United States of America 106, 19126– cattle rancher: identity with isolates found in conjunction with Borrelia 19131. burgdorferi and Babesia microti among naturally infected mice. Journal of Riess, T., Dietrich, F., Schmidt, K. V., Kaiser, P. O., Schwarz, H., Clinical Microbiology 37, 2598–2601. Schafer, A. and Kempf, V. A. J. (2008). Analysis of a novel insect cell Wilkinson, D. A., Duron, O., Cordonin, C., Gomard, Y., culture medium-based growth medium for Bartonella species. Applied Ramasindrazana, B., Mavingui, P., Goodman, S. M. and and Environmental Microbiology 74, 5224–5227. Tortosa, P. (2016). The bacteriome of bat flies (Nycteribiidae) from the Rinke, C., Schwientek, P., Sczyrba, A., Ivanova, N. N., Anderson, I. Malagasy region: a community shaped by host ecology, bacterial transmis- J., Cheng, J.-F., Darling, A., Malfatti, S., Swan, B. K., Gies, E. A., sion mode, and host-vector specificity. Applied and Environmental Dodsworth, J. A., Hedlund, B. P., Tsiamis, G., Sievert, S. M., Microbiology 82, 1778–1788. Liu, W.-T., Eisen, J. A., Hallam, S. J., Kyrpides, N. C., Woese, C. R. and Fox, G. E. (1977). Phylogenetic structure of Stepanauskas, R., Rubin, E. M., Hugenholtz, P. and Woyke, T. the prokaryotic domain: the primary kingdoms. Proceedings of the (2013). Insights into the phylogeny and coding potential of microbial National Academy of Sciences of the United States of America 74, dark matter. Nature 499, 431–437. 5088–5090. Rojas, N., Troyo, A., Castillo, D., Gutierrez, R., Harrus, S., Young, H. S., Dirzo, R., Helgen, K. M., McCauley, D. J., Billeter, S. Gutiérrez, R. and Harrus, S. (2015). Molecular detection of Bartonella A., Kosoy, M. Y., Osikowicz, L. M., Salkeld, D. J., Young, T. P. species in fleas collected from dogs and cats from Costa Rica. Vector- and Dittmar, K. (2014). Declines in large wildlife increase landscape- Borne and Zoonotic Diseases 15, 630–632. level prevalence of rodent-borne disease in Africa. Proceedings of the Roux, V. and Raoult, D. (1995). The 16S-23S rRNA intergenic spacer National Academy of Sciences of the United States of America 111, region of Bartonella (Rochalimaea) species is longer than usually described 7036–7041. in other bacteria. Gene 156, 107–111. Zeaiter, Z., Liang, Z. and Raoult, D. (2002). Genetic classification and Saw, J. H., Spang, A., Zaremba-Niedzwiedzka, K., Juzokaite, L., differentiation of Bartonella species based on comparison of partial ftsZ Dodsworth, J. A., Murugapiran, S. K., Colman, D. R., Takacs- gene sequences. Journal of Clinical Microbiology 40, 3641–3647.